Target States¶
BeeGFS uses reachability and consistency states for metadata nodes and storage targets. These states are important for the high-availability mechanisms in BeeGFS.
Display Target States¶
The reachability and consistency states can be displayed with the mode --listtargets
of the
beegfs-ctl
tool:
$ beegfs-ctl --listtargets --nodetype=storage --state
$ beegfs-ctl --listtargets --nodetype=meta --state
For more information on available parameters, please see the built-in help: beegfs-ctl
--listtargets --help
Description of Reachability States¶
The reachability state is used to describe whether the system can access a target. It is monitored by the management service. This state is an important part of the buddy mirroring, as failover of targets in a buddy mirror group is based on the reachability state of the primary target.
The following reachability states are defined:
Online: The target or node is reachable.
Probably Offline: A communication failure with this target or node was detected. This is an intermediate state, in which the information about a possible failure will be propagated to all nodes so that the system is prepared for the transition of this target to the offline state. If the unreachable target or node was shut down cleanly and comes back quickly (e.g., after a short service restart as part of a rolling update), or it was unreachable due to a temporary network failure, it can also get back to the online state and no failover or resync will happen.
Offline: The target is unreachable. If this is the primary target of a mirror buddy group, a failover will be issued immediately when the target transitions into the offline state.
Description of Consistency States¶
The consistency state describes the status of the data on a storage target or metadata node.
The following consistency states are defined:
Good: The data is considered to be good.
Needs resync: This consistency state applies only to secondary targets or nodes of a mirror buddy group. The data on the target or node is considered to be out-of-sync (e.g., because it was offline) and needs to be re-synchronized. Please see here for more information on re-synchronization possibilities: Synchronization of Storage Targets
Resyncing: The data on the target or node is considered to be out-of-sync, and a re-synchronization attempt is in progress.
Bad: The data on the target or node is considered to be out-of-sync, and a re-synchronization attempt failed. Please have a look at the logfiles for more information on why this state was selected.